Die deprocessing

[CTho9305]

I deprocessed (removed metal layers) a CPU die in hopes of getting die photos, and got decent results:
Pics (best one). Using this image as a reference, you can see the instruction cache, data cache, some of the L2 cache, and some of the blue structures in the "Fetch Scan Align Micro-code" section. Obviously the etch wasn't very controlled - towards the sides, almost nothing was removed, and towards the center it etched down to a very low metal layer.

It wasn't too complicated. I removed the lid with blades removed from a shaving razor and then an ordinary pocket knife. I got the die off the package with a blowtorch (which caused severe damage to the package - it was scorched, it bubbled a bit, it some of the surface mount components fell off, and a bunch of pins fell off...but there was no visible effect on the die itself) by heating it until it looked like flames were shooting out from under the die, and then pushing the die off (it didn't just fall off on its own). I etched it with glass etching cream (EXTREMELY DANGEROUS, contains hydrofluoric acid) ... I put some on, let it sit for a few hours, then cleaned the die off.

I think better results could be obtained by applying the acid more evenly, or by using liquid HF rather than a cream (but then you run the risk of splashing the acid and severely injuring yourself).

Many thanks to idontcare.

edit: fixed link

 

[Gibsons]

What molarity of HF is the cream you used?

 

[CTho9305]

Originally posted by: Gibsons
What molarity of HF is the cream you used?

I have no idea. It's consumer stuff (sold in crafts stores) labeled only as glass etching cream with warnings about how it can kill you.

 

[Idontcare]

Originally posted by: CTho9305
I deprocessed (removed metal layers) a CPU die in hopes of getting die photos, and got decent results:
Pics (best one).

Hey CTho9305 great pics man, awesome project. A little windex (ammonia) will cleanup the copper and make it nice and shiny again. (NH3 dissolve the CuOx)

Originally posted by: CTho9305
I think better results could be obtained by applying the acid more evenly, or by using liquid HF rather than a cream (but then you run the risk of splashing the acid and severely injuring yourself).

Get a plastic lid from something like a cool-whip container. (anything PFA or PTFE for HF compatibility) Spread the HF paste out on the lid to cover an area > the area of the die. The paste need not be very thick, couple mm's is good. Put the die onto the top of the HF paste (so now the die is upside down) and only press down slightly enough that you feel the die surface is covered. (don't squish the HF paste out the sides)

From there let gravity do the mechanical work (it will be even across the die surface) and let the HF paste do the chemical work. Poor man's CMP.

Also if you set your "processing chamber" (cool-wip lid) down inside your sink, again using gravity as your friend as a fail-safe that any splashes will naturally be working "uphill" to get out of the sink and onto you, you can add a smidgen of water to the paste to make it a little less viscous. From your pictures though it doesn't look like you need to speed up the process though.

 

[PottedMeat]

Wow thats pretty cool. My old solid state book has a micrograph of a PPC(I think thats what it was) with the silicon completely etched out. I've got a Radeon9800Pro BGA package here I could try that on.

Was there any adhesive/filler between the chip and the package?

Originally posted by: CTho9305
I deprocessed (removed metal layers) a CPU die in hopes of getting die photos, and got decent results:

Do you mean remove the Si/Si02 with the HF leaving the metal?

Can't really find much info on the strength of the etch cream, all I get is this MSDS:

http://shermanmillsartsupply.c...-GlassEtchingCream.pdf

30-30% Fluoride compounds by weight... and citric acid?

 

[Foxery]

This is why we don't buy you nice things!

What inspired this project?

 

[Idontcare]

Originally posted by: PottedMeat
Do you mean remove the Si/Si02 with the HF leaving the metal?

Can't really find much info on the strength of the etch cream, all I get is this MSDS:

http://shermanmillsartsupply.c...-GlassEtchingCream.pdf

30-30% Fluoride compounds by weight... and citric acid?

HF doesn't remove the Si nor the metal, but it removes the oxidized silicon (as in SiO2 and SiN). You need a much stronger oxidizer, such as nitric acid, to oxidize silicon and a metal as noble as copper.

 

[ColdFusion718]

HF is excellent at eating the calcium from your bones and killing you from the inside out. It's, however, considered a weak acid. Like Idontcare says, it's for etching oxides.

 

[silverpig]

<-- used 7:1 BOE today to etch oxide (the stuff that you need a splash apron, 2 gloves, a wetbench and some calcium gluconate cream handy in order to use).

I could probably do a nice job of it in the cleanroom at work.

 

[silverpig]

Originally posted by: ColdFusion718
HF is excellent at eating the calcium from your bones and killing you from the inside out. It's, however, considered a weak acid. Like Idontcare says, it's for etching oxides.

While it does eat the calcium from your bones, what actually kills you is when it takes the calcium from your blood. Apparently your heart needs some calcium in the blood in order to work.

When I first started working with the really nasty 49% HF stuff, I read some horror stories on the net about it. Some guy was working at a fumehood with 100 mL of 49% HF. It spilled on his legs and he was only wearing jeans. He took them off, his co-workers hosed him down and about 15 minutes later I think they applied some calcium gluconate gel. He went to the hospital. A few days later he lost a leg. A few days after that he died. 100 mL (that's a little over 1/3 cup).

 

[CTho9305]

Originally posted by: PottedMeat
Was there any adhesive/filler between the chip and the package?

Nope, just the solder bumps, which the heat from the blowtorch melted for me. Or, if there is one, it burned away well enough. Different package types may be more difficult to deal with though.

Can't really find much info on the strength of the etch cream, all I get is this MSDS:

http://shermanmillsartsupply.c...-GlassEtchingCream.pdf

30-30% Fluoride compounds by weight... and citric acid?

Interesting, thanks. You'd think they could have added a biological stain, or something to make it hurt, so it's blatantly obvious if you expose yourself.

Originally posted by: Foxery
This is why we don't buy you nice things!

What inspired this project?

Die photos look cool, what else?